All- Glass Staircase, Notting Hill, London Wilfried Laufs, Werner Sobek Ingenieure, Stuttgart, Germany To bring direct day- light into a domestic housing staircase area in London, an all- glass staircase has been built, where all treads are purely made of glass. The following article describes the design, detailing, calculation and construction of the staircase, which was also tested for sufficient strength and post- failure security aspects. Keywords: all- glass staircase, glass testing I. INTRODUCTION Staircases next to walls without windows might be dark and non- spectacular. If the staircase is yet to be constructed, why not use glass as the leading construction material, acting as the primary structural element. The following article describes such an all- glass staircase, which was designed and constructed in Notting Hill, London, in 2003 according to Figure 1. It is meant to give a practical example of the capacity of modern glazing, combined with structural as well as detailing knowledge within the field of structural glass engineering. II. STRUCTURAL CONCEPT AND DETAILING 2.1 Global Structural System To keep opaque structural material to a minimum, each tread was designed as an individual C- section, cantilevering out from the adjacent wall, where a hidden pair of steel beams spans across from concrete floor to concrete floor level with sufficient bending and torsional capacity, see Figure 2. Fig. 1: All- glass staircase Notting Hill, London 2.2 Glass Tread Profile Each C- profile is composed of three flat laminated glazing panels, rigidly glued to each other by means of acrylic bond, see Figure 3. For aesthetic clarity, the 90 corners of the C- section were chamfered to 45 at adjacent glass mitred joint edges before toughening and lamination. As internal temperature load cases are small ( T ~ 10 K only) and very little UV- light is hitting the intermediate bonding layers, acrylic bond was chosen as a suitable material instead of a PVBinterlayer. To achieve a good grip on each tread, non- slippery lines of ~ 0.5 mm depth were manufactured into the top surface (water jet). Page 1 / 6
2.3 Construction In order to avoid a direct steel- glass contact, but still transfer all support forces from the glass treads into the main steel support structure within the wall, a shoe- connection detail was developed according to Figure 4. A two- component resin was squeezed as the compatible intermediate mortar material between the glazing and mild steel flat profiles of the support (Figure 4), in addition to some distant- holding plastic support pads locally. Each glass tread was bonded together and fixed into its steel shoe. Each shoe was then bolted to the main steel beam in the wall, with options to adjust tolerances both vertically and horizontally. Neoprene pads (t = 3mm) were placed underneath each bolt to guarantee some spring behaviour against impact (abrupt steps); also shims can be added to align each tread in its exact final position. Fig. 2: Global structural system of glass stair (during construction) Fig. 3: All- glass tread C- section Fig. 4: Glass shoe support construction detail; all edges polished Page 2 / 6
2.4 Structural Calculations A load case according to Figure 5 was considered relevant, which two persons (100 kg each) crossing each other s way on the stair and stepping onto one tread at the same time, where a dynamic amplification factor of 1.5 was assumed. Fig. 5: Structural system and relevant loading per tread Assuming linear- elastic theory, the above loading would lead to a maximum un- factored upper tensile stress of max σ 1 = M/ W 1 yy upper = 2.13 knm / 390585 cm³ ~ 5.5 N/mm² (with full shear interaction of the bonding layers) and σ 2 = M/ W 2 yy lower = 2.13 knm / 57772 cm³ ~ 37 N/mm² (with no shear interaction) respectively, with σ allowable = 50 N/mm² > 37 > 5. One might expect the primary crack to start from the area of highest tensile stresses at the top, but as will be explained further down, this was not the case under high loads during every test. order of 120 to 200 N/mm² for short- term loading and would be expected at the area of highest tensile stresses. However, due to the unknown exact support condition (shoe with resin), where local pressure peaks or friction may occur, a 1:1 testing series was performed. A first glass tread with support shoe was tested, where the load P1 (see Figure 5) was increased in steps of 25 kg sand bags each (one bag per minute) up to failure (see Figure 6). The tip deflection at the free cantilever end was measured with results according to Figure 7. Due to a lack of budget, the shoe support itself was bolted to a non- rigid steel frame, simulating the wall behind, which also deflected under loading. Nevertheless, a rather linear loaddeflection curve was obtained, with the first crack coming from the centre of the left flange at 835 kg loading. The loading could be further increased, until the second glazing panel on the left flange side failed. The partially broken tread kept in place until at a 925 kg loading the system collapsed as a whole. 3.1 General III. TESTING As for most of modern glass constructions with primary load- carrying function, both strength and durability tests for regular usage as well as postfailure security tests for accidental cases need to be performed to satisfy all safety aspects and learn about the glass stair treads behaviour by means of 1:1 testing. 3.2 Ultimate limit state (static) Breakage of toughened glass usually is in the Fig. 6: Testing set- up with sand- bag loading for P 1 (top); primary crack at centre of inner laminate of flange (bottom) Page 3 / 6
Measured load - deflection curve (force- controlled) 1000 loading P 1 in [kg], speed v~ 1 min per 25 kg 900 800 700 600 500 400 300 200 100 left flange right flange one panel of leftt flange fails at 825 kg; system still stable for >30 minutes one panel of right flange fails at 950 kg; system still stable for >30 minutes second panel of right flange fails at 975 kg, total tread failure 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 relative cantilever end deflection f / f max [ - ] Fig. 7: Indicative force deformation behaviour of test rig with glass tread Test Breakage force Breakage stress comment [no] P 1 [kn] Full shear interaction [N/mm²] no shear interaction [N/mm²] 1 825 19 129 first crack 2 1250 29 196 at collapse 3 1125 26 176 at collapse Table 1: Theoretical breakage stresses However, as a system strength rather than a glass strength was tested within the steel shoe system test set- up, Table 2 cannot be taken directly for design, which is much rather achieved by 1:1 testing here, with 825 kg >> 2* P 1 = 300 kg for short- term ultimate loading conditions. 3.3 Ultimate limit state (dynamic impact) To examine the glass tread under possible abrupt high impact loads, a drop test was carried out using a 25 kg weight landing on the end of the tread and dropped from a height of 4 m. No glass or joints failed (Figure 8). Page 4 / 6
Fig. 9: Testing set up for cycling test Fig. 8: Drop test with a 25kg weight landing on the End of the tread, H ~ 4m, no failure 3.4 Serviceability limit state (long- term durability) As the glass strength might decrease with time and the steel shoe system needs to be durable, a cycle test was performed with the same test rig by loading the tread with 300 kg and measuring a tip deflection f 0, and mounting an electric motor with an eccentric cam which applied f 0, simulating the 300 kg load (approximately two people on one tread). The motor was left running 830 rounds per minute for approximately 10 hours (498000 cycles). This simulated an average family of 4, each using the stairs 4 times a day for 40 years (up and down). There was no breakdown of any of the joints, laminate or resin observed. Here, the acrylic bonding appears to be advantageous compared to the usual PVBinterlayer or resin products for laminate safety glazing: the fine broken glazing pieces stick to each other and still transfer compressive forces at the bottom of each flange for a long time and keep the C- section working under loading. Therefore, if one glass panel breaks, each laminate would stay in position long enough for a person to step on adjacent unbroken steps to be safe. In this context, it has to be added that the glass stair did not have to perform in case of fire, as there are other escape routes in the house. Therefore, fire resistance was not tested. 3.5 Post- Failure security To learn about remaining capacities of partially broken steps, both panels of one flange were broken on purpose by hammer. As shown in Figure 10 and Figure 11, the tread was still able to fully carry one person for at least 10 minutes. Fig. 10: Both panels of one flange broken, tread still carries one person Page 5 / 6
Fig. 11: Sufficient post- failure security observed IV. SUMMARY As shown above, a modern all- glass- stair construction is capable to carry high ultimate short- term loads as well as give a long- term durability for many years. Testing the system as a whole is recommended to find the true failure modes and learn from the broken system in terms of its post- failure capacity and breakage behaviour. In this case, even a partially broken tread can still carry a person long enough to walk down the staircase. The presented glass tread (Figure 12) represents a new generation of structural glazing applications and appears to be the first of its kind. Fig. 12: Finished glass tread and balustrade (not touching the treads) V. ACKNOWLEDGEMENTS Client: Chris Shirley, Notting Hill London Structural concept and Engineering: Whitbybird London, Wilfried Laufs / Will Stevens Construction and testing: Peter Collins, Hourglass Havant Hampshire Page 6 / 6